Table 1.
Number of samples assigned exclusively and significantly (p < 0.05 and p < 0.25, dependent correlations by Steiger[38]) to one of the four PCa subtypes, compared to the number of samples assigned when the four categories are combined into two categories with bad and good prognosis
|
A) Original signature | |||||||
|---|---|---|---|---|---|---|---|
| p-value threshold | BP-E/P/Pr | BP-ERG | GP1 | GP2 | Bad prognosis | Good prognosis | |
| PCa |
0.05 |
10 |
0 |
0 |
1 |
37 |
10 |
| |
0.25 |
22 |
3 |
2 |
11 |
52 |
21 |
| Normal |
0.05 |
0 |
0 |
2 |
7 |
0 |
31 |
| |
0.25 |
0 |
0 |
4 |
17 |
0 |
35 |
|
B) Signatures subtracted for the average normal signature | |||||||
| |
p-value threshold |
BP-E/P/Pr |
BP-ERG |
GP1 |
GP2 |
Bad prognosis |
Good prognosis |
| PCa |
0.05 |
9 |
1 |
0 |
1 |
53 |
4 |
| |
0.25 |
23 |
5 |
0 |
4 |
73 |
7 |
| Normal |
0.05 |
0 |
0 |
0 |
6 |
3 |
12 |
| 0.25 | 4 | 0 | 2 | 13 | 7 | 19 | |
While only few samples could be assigned exclusively to one of the four subtypes, a substantial number of samples could be assigned to one of the two combined categories of good and bad prognosis.